[FEA] allow Series.clip to clip accept Series as lower and upper bounds

[MarcoGorelli]

import cudf

data = {"a": [1, 2, 3, -4, 5], "lb": [3, 2, 1, 1, 1], "ub": [4, 4, 2, 2, 2]}
df = cudf.DataFrame(data)
df["a"].clip(df["lb"], df["ub"] + 1)

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-5-27928310299f> in <cell line: 5>()
      3 data = {"a": [1, 2, 3, -4, 5], "lb": [3, 2, 1, 1, 1], "ub": [4, 4, 2, 2, 2]}
      4 df = cudf.DataFrame(data)
----> 5 df["a"].clip(df["lb"], df["ub"] + 1)

/usr/local/lib/python3.10/dist-packages/cudf/utils/performance_tracking.py in wrapper(*args, **kwargs)
     49                     )
     50                 )
---> 51             return func(*args, **kwargs)
     52 
     53     return wrapper

/usr/local/lib/python3.10/dist-packages/cudf/core/indexed_frame.py in clip(self, lower, upper, axis, inplace)
   1004 
   1005         if len(lower) != self._num_columns:
-> 1006             raise ValueError(
   1007                 "Length of lower/upper should be equal to number of columns"
   1008             )

ValueError: Length of lower/upper should be equal to number of columns

expected (pandas output):

0    3
1    2
2    3
3    1
4    3
Name: a, dtype: int64

cudf version '25.02.00a225'

Could you please allow clipping by Series, like pandas does?

From your unofficial QA team at Narwhals

[mroeschke]

Thanks for the report.

Going to relabel as a feature request as this probably should be considered NotImplmentedError yet as the docs state

In case of Series/Index, lower is expected to be a scalar or an array of size 1

It appears the cuda kernel only supports cliping by a scalar value per column and would need to also support element-wise cliping by an equal length array-like